Probably might be, thank you for idea.
Interesting results in last two days; if I run Random Forest, it also get killed, but if I set parameter n_jobs = -1, it passed, and I got result after 50 min.
But again, if I use xgboost or its function xgb.cv() it also get killed.
Maybe I need to find how to limit memory usage?
Anyway, is there swan support which I can contact?
Hi Toni,
This is SWAN support, but your issue seems related to an excessive use of memory of the libraries you are using (since you said you also had the issue in your old laptop with limited memory). That is why I suggested to ask sklearn support about how to better manage memory. For anything else related to the SWAN service itself, please let us know if you have any issues.
Cheers,
Enric
Thank you Enric for your support, I really I appreciate it.
However I was wondering is it possible for me to limit memory use in SWAN, for let’s say, I want to fit data and want to limit memory use to 14GB so server won’t kill my processes.
Hi Toni,
SWAN allows you to set the memory limit when you start your session, but please note that does not set any limit on the application. This means that the limit is enforced by killing your process when it tries to go over it. I believe your best option is to learn how you can tweak that limit it from the application side.